A heartfelt message, estrogen replacement therapy: use it or lose it

Abstract

The issue of cardiovascular and cognitive health in women is complex. During the premenopausal phase of life, women have healthy blood pressure levels that are lower than those of age-matched men, and they have less cardiovascular disease. However, in the postmenopausal stage of life, blood pressure in women increases, and they are increasingly susceptible to cardiovascular disease, cognitive impairments, and dementia, exceeding the incidence in men. The major difference between pre- and postmenopausal women is the loss of estrogen. Thus, it seemed logical that postmenopausal estrogen replacement therapy, with or without progestin, generally referred to as menopausal hormone treatment (MHT), would prevent these adverse sequelae. However, despite initially promising results, a major randomized clinical trial refuted the benefits of MHT, leading to its falling from favor. However, reappraisal of this study in the framework of a “critical window,” or “timing hypothesis,” has changed our perspective on the benefit-to-risk ratio of MHT, and this review discusses the historical, current, and future approaches to MHT.

INTRODUCTION

It is well established that the risk of cardiovascular disease (CVD) in premenopausal women with intact ovaries is less than that of age-matched men (125). However, postmenopausal women have the same or greater risk of CVD as age-matched men (10). A major difference between pre- and postmenopausal women of similar age is the loss of ovarian hormones due to ovarian failure. Indeed, the levels of circulating estradiol in postmenopausal women is less than that in age-matched men (63, 83, 122). The earlier that ovarian failure occurs, either naturally or by surgical removal of the ovaries, the greater the risk of CVD (167). This association led to idea that estrogen replacement therapy (ERT), or estrogen plus progestin therapy (EPT), could sustain the CVD protection that functioning ovaries likely provide. The terms hormone replacement therapy (HRT) and menopausal hormone replacement therapy (MHT) have also been used to describe both ERT and EPT. To reduce ambiguity, this review will use ERT and EPT to describe estrogen only and estrogen plus progestin replacement therapy strategies and MHT where the specific nature of the hormone replacement therapy was not reported or was mixed. Throughout the review, we also attempt to specify the particular estrogens and progestins studied, since there are numerous types of these classes of compounds. For example, 17β-estradiol (E₂) is the major naturally occurring estrogen in women, whereas many studies of ERT have used conjugated equine estrogens (CEE) purified from horse serum, which contain 10 or more biologically active estrogens (23). At least two studies (148, 174) have reported that E₂ is superior to CEE in conferring cardiovascular and cognitive benefits.

The early observation that ERT and EPT substantially protected women from developing osteoporosis (6, 107) led to the wide embrace and use of ERT/EPT. The large-scale observational Nurses Health Study, which followed the health of 121,700 nurses for an extended period, suggested that both ERT (154) and EPT (55) also provided cardiovascular protective effects. In contrast, a concurrent study of 1,234 women (172) reported adverse cardiovascular effects of ERT. Conclusions from these studies were questioned, because these treatments were not documented in randomized clinical trials (RCTs) (70). Furthermore, they were criticized for having too many younger women and not enough older women (70). In addition, concerns for the “healthy woman bias” (5), whereby women who chose MHT tended to be healthier than women who chose not to use MHT, also could not be eliminated. To address this question, two RCT studies were initiated. The Heart and Estrogen/Progestin Replacement Study (HERS) (53, 70) investigated the effects of oral MHT on women with preexisting coronary heart disease. The Women’s Health Initiative (WHI) (92) studied the effect of EPT [oral administration of CEE plus medroxyprogesterone acetate (MPA) in healthy, mostly postmenopausal women] and oral CEE in women with hysterectomies. Both of these studies failed to show cardiovascular benefits of MHT.

Findings from these studies led to the recommendation that MHT not be used therapeutically for the prevention of CVD (9). Combined with concerns that estrogen can cause breast, ovarian, and uterine cancer, the use of MHT drastically declined (131). Further review of the WHI study, with stratification of the subjects by age, has revealed an interaction of age and effect of MHT (90). These analyses have led to a reconsideration of the potential benefits of MHT for protection against CVD in postmenopausal women in accord with “the timing hypothesis,” first put forward by Mikkola et al. (98) as “the window of therapeutic opportunity” and soon thereafter by Dubey et al. (40) and Mendelsohn and Karas (94). These observations were primarily derived from the dichotomous effects of ERT on atherosclerosis in monkeys associated with time of ovariectomy and time of initiation of ERT (3, 171). Mikkola et al. (98), however, also noted the disparity in ERT effects with age in the WHI study. Stated simply, the timing hypothesis [also known as the “window of opportunity” (1)] posits that initiation of ERT at or shortly after menopause confers cardiovascular benefits, whereas delayed initiation of ERT, more than 10 yr after menopause, has adverse cardiovascular effects. While the timing hypothesis has had its detractors (18), it is now generally accepted that initiation of ERT within 10 yr of either naturally occurring or surgically induced ovarian hormone deficiency confers cardiovascular benefits (66, 100), which complement its known benefits on osteoporosis without increasing overall cancer risk. An extension of the WHI, the Women's Health Initiative Memory Study (WHIMS), as well as other studies, including the Kronos Early Estrogen Prevention Study-Cognitive and Affective Study (KEEPS-Cog) (50) and Women’s Health Initiative Memory Study-Young (WHIMS-Y) (42), specifically directed to assessing the effects of MHT on cognitive function, now suggest that cognitive impairments associated with MHT in older (65–79 yr old) do not occur if MHT is initiated at ages of 50–54 yr (42) or within 6–36 mo of their last menses. Whereas KEEPS-Cog and WHIMS-Y, both of which used CEE, did not show cognitive benefits of MHT, other studies have suggested that early postmenopausal initiation of MHT can have neuroprotective effects on cognitive function and reduced risk of Alzheimer’s disease (AD) and other dementias (35, 99).

TAKE-HOME POINTS

Take-home points are as follows:

• 20th century studies that suggested beneficial effects of MHT on the cardiovascular system and cognition were likely correct despite the healthy women bias confounder

• Timing is critical for the initiation of MHT; the sooner it is initiated after menopause, the greater the benefits. But if initiation of MHT is delayed for more than 10 yr after menopause, it may cause more harm than benefit

• Estradiol generally confers greater benefits than CEEs

• Transdermal administration of estrogen affords greater benefits than oral administration

• Administration of estradiol immediately after menopause for a limited duration (<6 yr) may reduce the risk of breast and overall cancer incidence

• Unopposed estradiol (without added progestin) confers greater cardiovascular and cognitive benefits but should be used only in women who have had a hysterectomy

• Individual differences in risk factors for cancer and CVD should be evaluated as part of the criteria for determining whether MHT is appropriate

CURRENT UNDERSTANDING OF BLOOD PRESSURE AND CARDIOVASCULAR HEALTH

It is well recognized that treatment of hypertension is critical to preserving cardiovascular health. CVD, which is to a large extent secondary to hypertension, remains the number one killer in the United States and the rest of the world (108). The current guidelines for hypertension from the Joint National Committee on Prevention, Detection, Evaluation and Treatment of High Blood Pressure (JNC 7), released in November 2017 (169), recommends treatment for systolic blood pressure (BP) ≥ 130 mmHg. The change in guidelines is based on the recent SPRINT trial, in which cardiovascular outcomes for intensive treatment (≤120/80 mmHg) of individuals over 50 yr old with a high risk of CVD was highly beneficial. According to the old guidelines of beginning treatment in individuals with BPs ≥ 140/90 mmHg, the prevalence of hypertension is substantial in ~34% of the American population (https://www.heart.org/idc/groups/ahamah-public/@wcm/@sop/@smd/documents/downloadable/ucm_491265.pdf; accessed 1/8/18), and this percentage will greatly increase with the new lower guidelines.

OVARIAN HORMONES AND CARDIOVASCULAR HEALTH

Worldwide, women have lower BP than men: 122/77 vs. 127/79 (systolic/diastolic) mmHg, respectively (178a). Furthermore, CVD occurs earlier in men than in women. The incidence of a first cardiovascular event (per 1,000 person years in the 45- to 54-yr-old age group) is 10.1 in men and 4.2 in women (103). Menopause and oophorectomy are both associated with an increased incidence of hypertension and CVD (125, 136). Evidence from human and animal studies suggests that gonadal hormones contribute to these sex differences in the age of onset of hypertension and CVD and play a protective role in the cardiovascular health of women (12). Premenopausal women with functional ovaries that secrete estrogen have a lower risk of CVD than age-matched postmenopausal women (153), and early menopause (age: 40–45 yr) is associated with an increased risk of CVD (for a review, see Ref. 22). The functionality of the coronary endothelium is also healthier in young, premenopausal women than in older, postmenopausal women (93).

On the basis of these observations, it seemed logical that MHT would provide the cardiovascular benefits of intact functioning ovaries. In a meta-analysis of MHT studies, Grady et al. (54) found that the relative risk of coronary heart disease was 0.65 [95% confidence interval (CI): 0.59–0.71] in women who used ERT compared with women who never used estrogen. However, the WHI study (92, 130, 150), in more than 16,000 postmenopausal women, reported an increased risk of coronary heart disease in the EPT group over placebo controls [hazard ratio (HR): 1.29, 95% CI: 1.02–1.63] (92, 130) with no significant reduction in coronary heart disease with ERT (HR: 0.91, 95% CI: 0.75–1.12) (9). This finding plus the increased risk of stroke and pulmonary embolism in the ERT and EPT groups and the increase in invasive breast cancer in the EPT group led to early termination of the WHI study despite substantial reductions in colorectal cancer and hip fractures. Subsequent to this study, the use of MHT to provide cardiovascular benefit dramatically decreased (131).

In succeeding years, researchers have continued to monitor WHI subjects and have begun to challenge the conclusions of the WHI study, primarily based on the wide age range of the subjects, which masked age-related differential effects of MHT. Long-term observations and stratification of the different age groups now document that the time interval between menopause and initiation of MHT differentiates beneficial from adverse effects (24, 90, 98). This has led to the development of “the timing hypothesis” (64, 65), “critical period” (168), or “critical window theory” (87, 98, 170), which suggests that initiation of MHT soon after menopause has benefits that outweigh the risks, whereas delays, more than 10 yr postmenopause, in initiating MHT may have adverse cardiovascular and cognitive effects (40, 87, 101, 110, 132, 144, 170). The development of the timing hypothesis may be a classic case of déjà vu. In their 1991 paper describing a 10-year followup of the Nurses Health Study, Stampfer et al. (154) discussed a book chapter, written in 1987 by one of the authors of the 1985 Framingham study (172), stating that for coronary heart disease there was “…a nonsignificant protective effect among younger women, but a nonsignificant adverse effect among older women” [from Ref. 20 of Stampfer et al. (154)] from estrogen treatment. In the déjà vu all over again category, one of the original studies that reported beneficial effects of estrogen therapy for osteoporosis (6) reported that estrogen therapy begun 3 yr postoophorectomy increased bone mineral density, but when estrogen therapy was initiated 6 yr postoophorectomy, there was no effect on the rate of loss of bone mineral density. LaCroix et al. (80) reported an interaction between CEE therapy and age, with younger women showing a more favorable risk profile for myocardial infarction, total mortality, and global index of chronic diseases than older women.

We have used female Dahl salt-sensitive rats as a model for hypertension associated with salpingobilateral oophorectomy in women (152). In this model, rats were ovariectomized at either 4.5 or 7 mo and E₂ treatment was initiated at 7 mo. E₂ treatment reduced body weight and plasma angiotensin II in both groups compared with untreated rats; however, E₂ blocked the ovariectomy-induced increase in BP only in rats in which E₂ treatment was initiated immediately after ovariectomy. The expression of angiotensin II type 1 (AT₁) receptors was significantly increased in the brains of rats ovariectomized at 4.5 mo and treated with E₂ at 7 mo, whereas initiation of E₂ treatment immediately after ovariectomy prevented this increase in brain AT₁ receptor binding (152). These animal studies are consistent with the timing hypothesis, suggesting that a delay in initiating ERT after menopause reduces its efficacy and could have adverse effects on cognitive function. Interestingly, a study of women with elected oophorectomies without ERT revealed an increase in salt sensitivity in a subpopulation of those women, which could foreshadow development of hypertension later in life due to increased renin-angiotensin system activity (142). Tables 1 and 2 show the more contemporary view of the benefits and risks of MHT, when initiation of therapy occurs perimenopausally, less than 10 yr after menopause (approximately by an age of 59 yr old), or at later ages.

Table 1.

Cardiovascular and other risks associated with postmenopausal hormone replacement therapy (ERT/EPT)

Condition	Effect of ERT	Type of Estrogen	Reference(s)	Effect of EPT	Type of Estrogen, Progestin	Reference(s)
Coronary artery diseasea	↓Significant (30–55)*	CEE	55	↓Significant (30–55)*	CEE, MPA	55
	↓Significant (50–59 and 45–52)*	CEE, E2	24, 141	↓Significant (45–58)*	E2, NETA	141
	↓Moderate trend (50–55)*	CEE	159	↓Significant (55–79)*	CEE,norgestrel	159
	No change (57–98)	CEE	118	No change (55–80)	CEE, MPA	70
Cardiac function				↑Systolic and diastolic function (<3 yr postmenopause)*	Tibolone or CEE, MPA	160
Hypertension/reduction of BP	↓Nocturnal systolic, diastolic and mean BP, ↓day mean BP (56 ± 1.5)*	Transdermal E2,	143	↓Nocturnal systolic, diastolic, and mean BP (56 ± 1.5)*	Transdermal E2, vaginal progesterone	143
	↑Systolic BP (57–98)	Transdermal E2	20	↓Systolic and diastolic BP (mean age: 52.3 yr)*	E2, digesterone	163
	No change (40–70)	CEE	26, 28	↓Systolic and diastolic BP (47–55)*	E2, NETA	115
				Lower diastolic BP (45–53)*	E2/transdermal E2, CPA/progesterone	58
				No change or ↓trend (45–66)	U, U	125
Stroke/cerebrovascular event	Protection from ischemic stroke (<5 yr postmenopause)*	E2 or CEE	27	Increased risk (50–79)	CEE, norgestrel	159
	No change (50–59, 55.4 ± 2.4)	CEE	24	↑Large trend (50–79)	CEE, MPA	24
	No change (50–79)	CEE	159	No change (45–58)	E2, NETA	141
	No change (30–55) (45–58)	CEE	55, 141	No change (30–55)	CEE, MPA	55
Pulmonary or thromboembolism	↑Trend (50–79	CEE	24	↑Large (50–79)	CEE, MPA	24
				↑Large (55–80)	CEE, MPA	70
Hyperlipidemia/lipid profile/atherosclerosis progression	Improved profile (40–70)*	CEE, CEE/E2	26, 28, 161	Improved profile (45–64)*g	CEE, MPA/progesterone	161
	↓Progression (<6 yr postmenopause)*	E2	67	↓Progression (<6 yr postmenopause)*	E2, vaginal progesterone	67
				Improved profile (55–80)*	CEE, MPA	70
				↓Progression (<6 yr postmenopause)*	CEE/progesterone	140
Vasomotor symptoms	↓Large (42–79)*	CEE, E2	91, 140	↓Large (50–79)*	CEE, MPA	91
Type II diabetes	↓Small trend (50–79)*	CEE	24	↓Small trend (50–79)*	CEE, MPA	24
Obesity	No change (43–70)	E2, CEE	28	↓Small (45–58)*b	E2, norhisterone	74
	No changea	U	109	↓Trend (50–79)*	CEE, MPA	30
				No changea	U, U	109
Hip fracture	↓Moderate (50–79)*	CEE	24	↓Moderate (50–79)*	CEE, MPA	24, 130
Osteoporosisc	↓Moderate (54–72)*	E2	15	↓Moderate (50–79)*	E2/CEE/U, MPA/progesterone	15, 91, 130
Depressiond	Moderate improvement (50–56 and 40–55)e	Transdermal E2/CEE, transdermal E2	35a, 50	↓Incidence (45–60)*f	Transdermal E2, various	52
Cognitive impairment	Protective effect (<60)*	E2	68	Protective effect (<60)*	E2/CEE, MPA/various	68
	Reduction (66.4 ± 6.9)*	Raloxifene	176	Protective effect (PM-BSO)*	MPA	43
	Improved performance (66.2 ± 7)*	Raloxifene	73	Protective effect (menopause ≤ 5 yr)*	U, U	25, 144
	Improved performance (49–68)*	E2/CEE	174, 175	No change (menopause > 5 yr)	U, U	25, 144
	Impaired performance trend (65–79)	CEE	145	Impaired performance trend (65–79)	CEE, MPA	145
	No change (52.6 ± 2.6)	U	50	No change (<6 yr postmenopause)	U, U	61
				No change (52.6 ± 2.6)	E2/transdermal E2, progesterone	50
Alzheimer's disease and other dementias	Protective effect (40 and older)*	E2/transdermal E2/vaginal E2	99	Protective effect (40 and older)*	E2, MPA/NETA	99
	Protective effect (40–55)*	U	170	Protective effect*a,a	U, U	178
	Protective effect (28–94)*	U	76	Protective effect (“midlife”)*	Various, various	170
	Protective effect*a	U	113	↑Large (“late life”)	Various, various	170
	↓Risk (50–63)*	U	60	↑Incidence (65–79)	CEE, MPA	145, 146
	↓Risk trend (66.4 ± 6.9)*	Raloxifene	176	↑Moderate (50–79)	CEE, MPA	91
	↑Incidence (75–84)	CEE	145, 146
	↑Incidence (65–79)	CEE	170
	↑Trend (50–79)	CEE	91
Cardiovascular deaths	↓Trend (50–79)*	CEE	24, 90	↓Trend (50–79)*	CEE, MPA	24, 90
	↓Large trend (40–69 and 45–58)*	E2, CEE	26, 141	↑Trend (50–79)*	CEE, MPA	130
All-cause mortality	↓Moderate*a	CEE, E2, CEE	24, 90, 141, 159	↓Moderate trend (50–79)*	CEE, MPA, CEE, norgestrel	24, 90, 159
	↓Moderate trend* (45–79)	CEE, E2	24, 90, 141	No change (55–80)	CEE, MPA	70

Values in parentheses are ages of the subjects (in years). ERT, estrogen replacement therapy; EPT, estrogen-progestin therapy; CEE, conjugated equine estrogens; E₂, 17β-estradiol; MPA, medroxyprogesterone acetate; NETA, norethisterone acetate; CPA, cyproterone acetate; U, unspecified; PM-BSO, premenopausal bilateral salpingo-oophorectomy onset menopausal hormone therapy (MHT).

^*Beneficial effects.

^aUnspecified age range, generally from meta-analysis data; includes myocardial infarction and heart failure.

^bThese studies reported a redistribution of fat mass to the abdomen (increased waist circumference and waist-to-hip ratio) as well as a shift from lean to fat mass and greater loss of bone mineral density in women who did not receive MHT; such changes are associated with an increased risk of cardiovscular disease (30, 81).

^cStudies that showed increased bone density as a surrogate for reduced osteoporosis.

^dDepression has been shown to be a risk factor for cardiovscular disease, in particular ischemic heart disease (16, 49, 162)

^eFour-year followup only; improvement in depression was seen only with oral conjugated estrogens and nontransdermally administered estrogen; may include some individuals with EPT.

^fImprovement was limited to women in early menopause transition.

^gCyclic progestin was better than continuous progestin.

^hThe protective effect arose from early postmenopausal hormone replacement therapy use or usage for more than 10 years.

Table 2.

Cancer risk associated with hormone replacement therapy (ERT/EPT)

Condition	Effect of ERT	ERT	Reference(s)	Effect of EPT	EPT Composition	Reference(s)
Breast cancer	↓Moderate trend (45–58)*	E2	141	↓Moderate trend (45–58)*b	E2, NETA	141
	↓Small trend (50–79)*	CEE	24, 91	↑Moderate trend (50–79)*	E2 or CEE, MPA	24, 69, 70, 130
	↓Significant (60–69)*	CEE	8	↑Moderatea	U	159
	↓Significant (50–79)*	CEE	80	↑Large (50–55)	Various	34
	No change (50–55)	U	159	↑Largea	U	165
	↑Smalla	U	162
Ovarian cancer	↑Moderate (50–74)a	U	62, 117	↑Small (50–74)a	U, MPA, NETA	62, 117
Endometrial cancer	↑Smalla	Various	54, 116	↓Moderate trend*	CEE, MPA,	70, 116
Cervical cancer	↑Insignificant (35–70)	U	134	No change (35–70)	U	134
Colorectal cancer	↓Moderate trend (50–79)*	CEE	24, 130	↓Moderate trend*	CEE, MPA	24
				↓Moderate (50–79)*	CEE, MPA
All cancers	↓Itrend (50–59)*	CEE	24	No change (50–59)	CEE, MPA	24
All cancer deaths	↓Moderate trend (50–59)*	CEE	24	↓Moderate trend (50–79)*	CEE, MPA	24, 70, 90
	↓Trend (50–79)*	CEE	86, 90

Values in parentheses are ages of the subjects (in years). ERT, estrogen replacement therapy; EPT, estrogen-progestin replacement therapy; NETA, norethisterone acetate; CEE, conjugated equine estrogens; MPA, medroxyprogesterone acetate; U, unspecified; E₂, 17β-estradiol.

^*Beneficial effects.

^aUnspecified age range, generally from meta-analysis data.

^bDecrease was greatest in younger women (<50 yr old) (69); reported wide variations in breast cancer risk based on ethnicity, breast density, and obesity (24); limited to an age group of 50–59 yr old, also includes reanalysis of the age cohort of 50–59 yr old of the Women's Health Initiative study with generally similar results (62); 5 yr posthormone replacement therapy, no increase in ovarian cancer risk was present.

MHT AND CANCER

The incidence of reproductive tissue cancer in women generates a considerable amount of fear. Thus, any report of an increase in reproductive cancers in women leads to a reflexive avoidance of the causal factors no matter how small the increase in risk. Table 2 shows cancer risks of MHT with a primary emphasis on the risks associated with perimenopausal or early postmenopausal initiation of MHT (<10 yr postmenopause) initiation of MHT or at later ages. Of note, LaCroix et al. (80) also reported an interaction between age and colorectal cancer, with younger women showing lower risk than older women, consistent with the timing hypothesis.

Relative Risk Versus Absolute Risk

An important consideration when balancing risk-to-benefit ratios is not only the increase in relative risk of disease but also the incidence of the disease. Santen and Petroni (139) referred to this as “relative vs. attributable risk” and point out how misleading relative risk determinations can be. They determined that the attributable risk of a woman getting breast cancer arising from taking ERT for 10 yr starting at an age of 50 yr old was 2.8 per 100 women but that 11.9 per 100 women would be spared an adverse cardiovascular event, resulting in a net reduction of 9.1 incidences of disease (Fig. 1). Since not everyone who develops breast cancer or CVD will die from it, the numbers of people who die from breast cancer decreases to 0.67 per 100 women, while the number of people saved from CVD-associated deaths is 4.4 per 100 women (139). Furthermore, if the reduction in the relative risk of women dying from hip fractures with EPT is added in, the benefit-to-risk ratio becomes even greater. Moreover, a reappraisal of the effects of ERT in women with hysterectomies in the WHI suggests that there may be no increase, and a possible decrease, in the incidence of breast cancer with ERT (9, 155). This factor was also noted in a statement of the Endocrine Society supporting the use of “menopausal hormone therapy” in women aged 50–59 yr old or within 10 yr of menopause (138).

Fig. 1.

Relative and attributable risks for breast cancer and cardiovascular death with estrogen replacement therapy (ERT). The incidence of breast cancer is based on statistics from the National Cancer Institute website (https://www.cancer.gov/types/breast/risk-fact-sheet; accessed August 6, 2018). The incidence of cardiovascular disease deaths was derived from Lloyd-Jones et al. (83). The estimate of the relative risk for breast cancer and coronary heart disease with ERT was from Santen and Petroni (136) but can vary considerably based on the type of ERT agents, routes of administration, duration of therapy, followup period, and individual risk factors.

It should be kept in mind, however, that individual risks for breast cancer and CVD can also vary and can alter the risk-to-benefit ratio, e.g., a person at high risk for breast cancer and minimal risk for CVD might be adversely affected by EPT. However, a person with minimal risk factors for breast cancer and at high risk for CVD might gain substantially from EPT. More recent studies, LaCroix et al. (80) and Gurney et al. (56), suggested that the relative risk of breast cancer was reduced when ERT was initiated within the critical window postmenopause and continued for an average of 5.9 yr.

Although not directly applicable to MHT cancer risks, a recent study of birth control pill (estrogen + progestin) usage in 1.8 million premenopausal women concluded that the risk for breast cancer was higher for women using hormonal contraception compared with nonusers (102). However, another study of the use of oral contraceptives indicated reduced risks for colorectal, endometrial, ovarian, lymphatic, and hematopoietic cancers in women taking birth control pills, concluding that the decrease in select cancers balances out and may override the increase in breast and cervical cancer incidence (72, 101). An analysis of the net effect of oral contraceptive usage in England concluded that their usage prevented 1,600 cases of cancer in 2010 (116).

Variations in MHT

Analysis of the risk-to-benefit ratio of postmenopausal MHT is complicated by many variables: unopposed estrogen versus estrogen plus progestin (with either continuous or cycled progestin), idiosyncratic characteristics of the estrogenic and progestogenic agents (cf. Refs. 31, 40, 88, 148, 168, and 174), estrogen receptor (ER) subtype selectivity, route of administration of MHT, MHT dosages, and duration of MHT. Kuhl (79) provided a comprehensive review of the variables associated with different estrogenic and progestogenic preparations, their routes of administration, relative potencies, pharmacokinetic characteristics, and tissue selectivity that covered these issues in much greater detail than is possible in this brief review. In general, oral bioavailability of estrogens is poor due to limited absorption and metabolism in the gastrointestinal tract and liver; however, transdermal absorption of estradiol varies widely between individuals. A recent meta-analysis reported that oral, but not transdermal, administration of estradiol is associated with an increased risk of thromboembolism (135). In addition, oral (58, 75), but not transdermal (58), estradiol increases angiotensinogen production by the liver. Although the resulting increase in plasma angiotensin II does not appear to increase BP, it is noteworthy that a study comparing oral and transdermal administration of estradiol found a small reduction in BP in normotensive oophorectomized women with transdermal, but not oral, administration (7). Manhem et al. (89) also found a small reduction in BP in hypertensive postmenopausal women with transdermally administered estradiol. Additionally, after 12 mo of women receiving transdermal estrogen, Beljic et al. (20) observed that postmenopausal women with normal BP saw their systolic BPs significantly decrease, as did Seely et al. (143), whose study of 15 normotensive women showed a significant decrease in diastolic and mean BP (~5 mmHg) with the administration of transdermal estradiol. Thus, while MHT generally does not reduce the postmenopausal rise in BP (123, 125) and may increase BP compared with placebo-treated women (158) regardless of time past menopause, this may be a result of a preponderance of studies in which the estrogen was administered orally. In view of the studies in which transdermally administered estradiol caused small to moderate reductions in BP (7, 58, 89, 143), one of which concurrently showed no change in BP with oral administration of estradiol (7), as well as two studies with orally administered estradiol plus one study of combined oral estradiol and progestin that have shown reductions in BP compared with placebo controls (58, 75, 163), the ability of transdermally administered estradiol to lower BP should be investigated on a larger scale.

Additional variables that should also be considered are smoking/tobacco usage, which is generally controlled for in studies of ERT and EPT but which is associated with substantially increased CVD and incidence of cancer (13, 86, 126, 157); lifestyle (124, 126); hysterectomy; age at menopause or surgically induced menopause (see below); ethnicity (112, 157); individual differences (79); lipid profile; body weight/body mass index (34) and metabolic syndrome, which is associated with increased cardiovascular risk factors such as increased BP, dyslipidemia, type 2 diabetes mellitus (4), and lean versus obese women (137); duration of MHT therapy; duration of followup after MHT; weighting of adverse versus beneficial outcomes, e.g., increased risk of specific cancers versus reduced risk of other cancers and osteoporosis; and what is now most apparent, the time between menopause and beginning MHT. Of note, women who never smoked were more likely to have a later age at menopause than current tobacco smokers (39, 82). Moreover, as reviewed by Appelman et al. (10), women smokers have a greater increase in their CVD risk than men.

Thus, although it is still not possible to make a definitive generic determination in favor of MHT for cardiovascular benefits, newer data are increasingly in favor of MHT when initiated perimenopausally or within the critical window postmenopausally (36). In particular, in women who have premature ovarian insufficiency before an age of 40 yr old, either naturally or surgically, have an increased risk of early mortality, CVD, and neurological disease (147), and MHT is strongly recommended unless there are contraindications (14). This population has higher BP than age-matched women who are premenopausal (for a review, see Ref. 136). Additionally, MHT is well established for relief of perimenopausal vasomotor symptoms: hot flashes and night sweats, sleep disturbances, as well as urogenital atrophy (14, 32, 43, 140). In addition, the use of ERT in women who have had a hysterectomy appears to have more benefit than EPT in women with an intact uterus (36, 54, 90, 101, 141).

One issue related to postmenopausal MHT that has not been studied is cyclic administration of an estrogen. In premenopausal women, estrogen as well as progesterone levels vary widely over the menstrual cycle (173), whereas postmenopausal estrogen therapy and many EPT regimens maintain a constant level of estrogen and progestin. Some MHT regimens cycle the progestin, most often 12 days on, 18 days off, which is associated with a greater increase in high-density lipoprotein-cholesterol (HDL-C) than MHT with continuous progestin (101a). There is anecdotal evidence of cyclic postmenopausal use of ERT by some women, but we found only one study, a short-term, randomized controlled trial, of 54 postmenopausal women (age: 50–65 yr) which used cyclic estrogen (4) according to its use as a birth control pill, reporting an adverse effect of EPT on HbA_1c levels in “normoglycemic” (HbA_1c < 8%) women with type 2 diabetes mellitus. This study contrasts with an observational study of 15,435 women with type 2 diabetes mellitus reporting that MHT use was associated with lower HbA_1c levels (45), which was challenged as being subject to the “healthy women bias” (5).

INFLUENCE OF ER SUBTYPES

The issue of ER subtype selectivity and expression may play a role in the cardiovascular and cognitive protection afforded by ERT. There are three known ER subtypes: ERα, ERβ (33), and G protein-coupled ER (GPER), also known as GPR30 and GPER-1 (44, 97). The classical ERs, ERα and ERβ, are cytoplasmic receptors that become transcription factors when bound by estrogen, thereby regulating the expression of a number of genes, although ERα (29) and ERβ can also localize to the plasma membrane and signal through protein kinase pathways (for a review, see Ref. 114). Acute activation of ERα also has nongenomic effects such as activating endothelial nitric oxide (NO) synthase (40). Consistent with either GPR30 or membrane-localized ERα, estradiol was shown to induce a short-latency NO release from cultured endothelial cells (156). NO induces vasodilation, dilating blood vessels by relaxing the smooth muscles lining the vessels. ERα has a significant role in maintaining “cardiometabolic health” (33), and its mRNA has been shown to decrease with aging in mouse aorta (111), although its protein expression remains relatively constant in the human uterine arteries (110).

Animal studies of ERβ have also indicated largely beneficial actions on the cardiovascular system, although ERβ is also associated with adverse effects. A study of uterine ERβ expression showed that it increased with time past menopause, corresponding to an increased inflammatory response to estrogen (110). ERT increased NO production and decreased NADPH oxidase activity in young but not old mice, concomitant with an increase in the ERβ-to-ERα ratio in the thoracic aorta of old versus young mice (111). However, a recent review of human and animal studies suggested that ERβ exerts a host of protective actions in the cardiovascular system, primarily targeting vascular endothelial cells (105).

GPER activation has a direct vasodilatory effect (11, 44, 57) and both activates (57) and inhibits (44) cAMP formation. It also transiently activates MAPKs via transactivation of the epidermal growth factor receptor (46) and other signaling pathways (44) in response to estrogen. It is present in osteoblasts (59) and may be a major contributor to the osteogenic effects of estrogen (78). Of note, GPER is also a receptor for aldosterone, for which it displays a higher affinity than for estrogen (44).

ADVERSE EFFECTS OF ESTROGEN LOSS, HYPERTENSION, AND CVD ON COGNITION AND PROGRESSION TO DEMENTIA

As noted above, high BP, at least during midlife, is a risk factor for cognitive impairment and dementia (for a review, see Ref. 77). And, while MHT at best lowers BP slightly, its beneficial effects on vascular endothelium may have the same cardiovascular protective effects on the cerebral microvasculature as a reduction in BP. Patients with AD with hypertension show a faster rate of cognitive decline than normotensive patients (21). Furthermore, patients with untreated hypertension show a more rapid progression from mild cognitive impairment to AD than patients whose hypertension is treated with a variety of antihypertensive agents (84). With additional therapy for diabetes and hyperlipidemia, the progression to AD was slowed even more (63). Similarly, patients showing good cardiovascular health in terms of BP and blood glucose have more favorable cognitive function later in life (51). Recent studies in humans have suggested there is a link between endothelial dysfunction and AD pathology (37). Geriatric patients with masked hypertension had significantly lower scores on Mini-Mental State Exam test and Categorical Fluency Test than normotensive patients (41). A large meta-analysis of antihypertensive drug therapies shows convincingly that reduction of elevated BP is associated with a substantial reduction in the risk of dementia (133). In a geriatric African-American cohort, cognitive decline was decreased by 38% with antihypertensive therapy, independently of the drug class (106). Interestingly, an increased risk of cognitive decline in women with elevated systolic BP and pulse pressure has been reported (177). Finally, a meta-analysis of 18 studies concluded that there was a strong trend toward decreased cognitive impairment with antihypertensive therapy, with Ca²⁺ channel blockers and renin-angiotensin inhibitors showing the greatest benefits (133).

With respect to the effects of estrogen on cognitive function and dementia, especially AD, it has long been known that estrogen has neuroprotective effects (95, 180) (for reviews, see also Refs. 38, 47, 48, 96, 120, and 129). These positive neurotropic effects have been reported to be mediated largely by the ERβ receptor subtype (for a review, see Ref. 164). However, in the hippocampus, the brain region most closely associated with memory and learning, estradiol is reported to act on both ERα and ERβ as well as on GPER to promote synaptic plasticity and neuroprotection (151, 166, 179); thus, all of these receptor subtypes likely mediate the improvement of memory in rodent models of cognitive function. An early uncontrolled study of the effects of ERT in women of unspecified ages in a retirement community showed a reduction in risk of developing AD, with the reduction increasing with dose and duration of treatment (113). Although the randomized controlled WHIMS initially suggested an adverse effect of MHT (with CEE as the estrogenic agent) on cognitive function and AD (42, 145), a recent 18-yr followup study of the WHI study (90) revealed that the CEE-only group had a significantly lower AD mortality (HR: 0.74, 95% CI: 0.59–0.94). The Cache County, UT, study of more than 1,800 women showed that MHT reduced the risk of AD, with the longest usage (>10 yr) having the lowest risk (HR: 0.41, 95% CI: 0.17–0.86) (178). Another recent long-term study (71) showed that long-term (>10 yr) but not short-term (<10 yr) MHT was associated with a reduced risk of developing AD (HR: 0.53, 95% CI: 0.31–0.91). Interestingly, a recent study (19) to determine whether the timing hypothesis applied to estradiol’s beneficial effects on cognition in the nonhuman primate model in which the timing hypothesis for estradiol’s cardiovascular benefits occurred indicated that a much longer duration between menopause and ERT could elapse, during which estradiol could benefit cognitive function. This observation is echoed in a large-scale, 489,105 women, study of MHT, which showed a reduction in vascular dementia independent of time of initiation of MHT (99). Of note, those investigators (99) suggested that the improved cardiovascular risk profile of E₂ compared with CEE (141, 148) might explain the large (37–39%) reductions in the risk of death from vascular dementia in their cohort. Additionally, Mikkola et al. (99) also reported a risk reduction for AD with longer duration, >5 yr, of ERT/EPT. Studies reviewed by Faubion et al. (43) also support beneficial effects on cognitive function and reduction of risk of dementia with MHT in accord with the timing hypothesis.

The selective ER modulator (SERM) raloxifene, which is a weak partial agonist of ERα with negligible agonist effect on ERβ (17), and a preferential estrogenic agent on bone, was shown to significantly reduce the risk of mild cognitive impairment (RR: 0.67, 95% CI: 0.46–0.98) and marginally reduce the risk of AD in a large cohort (7,487 women) with an average age of 66.3 yr old (176). In another study conducted on women aged 70–80 yr old, raloxifene improved verbal memory compared with a placebo group (73), which suggests that the cognitive benefits of SERM usage may not be subject to the timing hypothesis effect.

Additionally, as with CVD, the postmenopausal population that has undergone premenopausal oophorectomy or early menopause appears to be at higher risk for cognitive impairment and dementia (14, 35, 104, 147) than those who have undergone natural menopause. The Mayo Clinic Cohort Study of Oophorectomy and Aging showed that the risk of cognitive impairment or dementia nearly doubled in women who underwent bilateral oophorectomy before the age of natural menopause regardless of the indication for oophorectomy (e.g., benign ovarian condition or for cancer prophylaxis) (127). This first large-scale study of cognition with long-term followup also showed that the risk of cognitive impairment and dementia increased the younger the age at oophorectomy. These findings from women living in the United States were confirmed in a Danish historical cohort study, published in 2010, that queried national disease registries (119). The Danish study showed that the risk of dementia with onset before the age of 50 yr increased in women who underwent bilateral oophorectomy before the age of natural menopause, again with the risk increasing the younger the age at surgery. Furthermore, two longitudinal studies analyzed by Bove et al. (25) showed that women who underwent surgical menopause at younger ages saw more rapid rates of cognitive decline and found that the effect of each year of earlier surgical menopause matched the effect of 6 mo of aging in relation to the rate of cognitive decline. However, when administered within the 5-yr “window of opportunity,” MHT significantly slowed cognitive decline and risk of AD, in contrast to a worsening of cognitive decline and risk of AD with delayed initiation of MHT after an age of 65 yr (for a review, see Ref. 128).

CURRENT GAPS IN KNOWLEDGE

The conflicting results from studies of MHT are likely due to differences in experimental design, including the composition of the therapeutic agents (e.g., type of estrogen and progestin), the mode of treatment (e.g., oral vs. transdermal or vaginal), protocol (e.g., continuous or cycling progestin), treatment duration and time of initiation postmenopausally, type of menopause, outcomes, and followup periods. In addition, variations in expression and functionality of the different ER subtypes complicates the evaluation of the effects of postmenopausal MHT. These countless permutations currently limit our ability to compare study outcomes on cardiovascular and cognitive health as well as cancer risk and other effects. Individual differences in genetics and environment are also factors that differentially affect a woman's response to MHT. Thus, even when the ideal MHT regimen for one's health is identified, personalized medicine will remain critical to determining whether or not MHT is indicated.

LIMITATIONS OF THE STUDY

Entering the search term menopausal hormone therapy in PubMed yielded 35,924 articles. In selecting 180 articles for this review, we have cited ~0.5% of the literature on this topic. As a result, there is considerable potential for unintentional bias and omission of important studies despite our best efforts to focus on the most salient articles relevant to our goal of evaluating the relative benefits and risks of MHT. In addition, the possibility of publication bias in favor of studies showing significant effects cannot be ignored, although this bias would favor both positive and negative effect observations similarly.

SUMMARY AND CONCLUSIONS

It is increasingly apparent that postmenopausal MHT has cardiovascular and cognitive benefits, provided it is initiated within the critical window of opportunity, i.e., <10 yr postmenopause, and it has even better outcomes if 1) it is initiated perimenopausally, 2) it is administered transdermally, and 3) the estrogenic agent is E₂. Moreover, hysterectomized women appear to achieve better overall health outcomes from ERT than women with intact uteri treated with EPT. The adverse effects of delayed MHT, which distorted the results of the WHI study, have now been stratified from the early MHT treatment, further affirming the benefits of early MHT. Moreover, the breast cancer risk of MHT is far less than originally reported in the WHI study, with a reduced risk in women with hysterectomies treated in the early postmenopausal period only with estrogens for <6 yr (56, 80). Added to the reductions in colorectal and other cancers associated with MHT, there is a net reduction in all cancers as well as all-cause mortality with MHT. However, despite these positive observations, recommendations for MHT are still limited to prevention of osteoporosis and perimenopausal vasomotor symptoms or protection from adverse cognitive effects of early or premenopausal surgical menopause (108a).

Lingering issues yet to be resolved are how long should MHT be continued and whether it should be ERT or EPT. In view of the study suggesting that MHT of >10 yr in duration provides better protection against AD than <10 yr in duration (178) and the beneficial effects of continued MHT on mood, the possibility of lifelong MHT cannot be dismissed. There is evidence that more women are continuing to take MHT for more than a decade, so observational studies will help determine the optimal duration for MHT, although the “healthy women bias” may continue to be a confounder in interpreting such studies. With close monitoring and continuing assessment of genetic and environmental risk factors for cancers, dementia, and CVD, there can be a refinement of the criteria for determining the optimal use of MHT.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grant R01-HL-119380 (to R. C. Speth, H. Ji, and K. Sandberg), the Peptide Radioiodination Shared Resource, Georgetown University (to R. C. Speth), and the Cardiovascular Neuroscience Fund, Nova Southeastern University (to R. C. Speth).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.C.S. and K.S. conceived and designed research; R.C.S., H.J., and K.S. performed experiments; R.C.S., H.J., and K.S. analyzed data; R.C.S., H.J., and K.S. interpreted results of experiments; R.C.S. and H.J. prepared figures; R.C.S., M.D., and K.S. drafted manuscript; R.C.S., M.D., H.J., and K.S. edited and revised manuscript; R.C.S., M.D., H.J., and K.S. approved final version of manuscript.